Old-growth coast redwood forests have been significantly altered by logging activities, resulting in highly fragmented second-growth stands throughout their range.
Although these second-growth forests have made a rebound from the highly destructive logging practices of the 19th and 20th centuries, the recovery of this forest system as a whole is not well understood.
Historically, clear-cutting was the predominant method of timber extraction in coast redwood forests. In recent years, selective-harvest has become more common and is currently the only harvest method employed in Santa Cruz County where clear-cutting is no longer permitted. Although this practice is generally considered less damaging, it still may have lasting impacts on forest composition and structure.
16 Research Objective
The objective of this study was to determine how the composition and structure of
selectively-harvested coast redwood stands were influenced by the following independent variables: years since last harvest, number of harvest re-entries and the percentage cut per hectare.
Research Questions
1. Which independent variable had the strongest influence on canopy cover, stand density and density of large woody debris (LWD)?
2. Which independent variable had the strongest influence on basal area and dominance of coast redwood?
3. Which independent variable had the strongest influence on floristic composition, including coast redwood associated species and exotic species?
Hypotheses
The following predictions were made in regard to the research questions above:
H1 The percentage cut per hectare would be the strongest independent variable for canopy cover, stand density and density of LWD.
H2 Basal area and dominance of Sequoia sempervirens would decrease in relation to the number of harvests and percentage harvested, but would increase in relation to years since last harvest.
H3 The percentage cut per hectare would be the strongest independent variable for floristic composition, including coast redwood associated species and exotic species.
17 METHODS Study System
Santa Cruz County has a Mediterranean climate typical of California’s coast, defined by hot, dry summers and high precipitation during the winter. Annual rainfall is between 100-150 cm and morning fog is common during the summer months.
Common coast redwood associate hardwoods in this region include California hazelnut (Corylus cornuta), California box elder (Acer negundo), California bay, Pacific madrone, big leaf maple (Acer macrophyllum), coast live oak (Quercus agrifolia) and interior live oak (Quercus wislizeni) (Cooney-Lazaneo and Lyons 1981; Lyons and Cuneo-Lazaneo 1988). Common coniferous associates include California nutmeg (Torreya californica), Douglas-fir and knobcone pine (Pinus attenuata).
There are a variety of avian, mammalian and amphibian species that occupy Santa Cruz County. Birds found in this habitat include the Cooper’s hawk (Accipiter cooperii), golden eagle (Aquila chrysaetos), long-eared owl (Asio otus), marbled murrelet,
peregrine falcon (Falco peregrinus), osprey (Pandion haliaetus), sharp-shinned hawk (Accipiter striatus) and yellow warbler (Dendroica petechial) (California Department of Forestry and Fire Protection 2001). Mammals within the region include the dusky-footed woodrat (Neotoma fuscipes), Mexican black-tailed deer (Odocoileus hemionus),
mountain lion (Puma concolor), raccoon (Procyon lotor) and deer mouse (Peromyscus maniculatus). Amphibian species present in this range include the California red-legged frog (Rana aurora draytonii), foothill yellow-legged frog (Rana boylii) and southwestern
18
pond turtle (Actinemys marmorata pallida) (California Department of Forestry and Fire Protection 2001; Bulger et al. 2003).
Selective-harvest has been the required method of timber extraction in Santa Cruz County since clear-cutting was banned in 1970; this was in adherence to specific
regulations applicable to the Southern Subdistrict under the California Forest Practice Rules (California Department of Forestry and Fire Protection 2013). The coast redwood forests in this region are predominately second-growth with sparse patches of old-growth occurring intermittently throughout the county.
The study sites for this research included five different harvests within the Byrne-Milliron Forest, located in Santa Cruz County (Figure 2). The forest is located five miles north of Corralitos and is composed of coast redwood, mixed chaparral, montane
hardwood/coastal scrub and eucalyptus. Clear-cut in the late 1880s, the property predominately consists of even-aged coast redwood and Douglas-fir (California Department of Forestry and Fire Protection 2001).
19 Figure 2. Study system. Image by Derrick Wynes.
The preserve spans 163 hectares with elevations ranging between 125-500 m and predominately west facing slopes (Overtree and Kitayama 2013). Situated atop the Purisima Formation, the soil types on the preserve include the Ben Lomond-Felton complex, Lompico-Felton complex, Nisene-Aptos complex and a small portion of
Pfeiffer gravelly sandy loam located near the entrance (California Department of Forestry and Fire Protection 2001). Several small tributaries on the property lead to the nearby Browns Creek, although only one runs continuously throughout the year.
Named for its original owner Carlton Byrne, the preserve was acquired by the Land Trust of Santa Cruz County in 1984. The Land Trust initially only purchased the Byrne property, which consisted of 130 hectares. In 2007, the Milliron property, an additional 32 hectares, was acquired through a conservation easement and then later
20
purchased in 2008 (Overtree and Kitayama 2013). For the purpose of this study the Milliron property has been excluded.
There have been a total of seven selective harvests on the preserve since the Land Trust took ownership. The property is now broken up into several management units (Figure 3) including the Central Unit (56 hectares), the Early Successional Unit (22 hectares), the Late Successional Unit (15 hectares) and the Southern Unit (24 hectares).
Many of these units have undergone varying levels of cutting intensity implemented with the use of either tractor, skyline cable, or both methods combined (Table 1).
Figure 3. Management units at Byrne-Milliron Forest. Image by Derrick Wynes.
21 Table 1. Harvest details among study sites.
1987/2004 1990/2007 1996 2001 2007
Five of the selective-harvests have occurred in the Central Unit; two of these sites were harvested twice. The first site was harvested in 1987 and re-entered in 2004 and the second site was harvested in 1990 and re-entered in 2007. Both initial harvests were completed under Timber Management Plans (THP) and the latter were completed under one Non-Industrial Timber Management Plan (NTMP) prepared in 1991 (Overtree and Kitayama 2013). All harvests after 1996 were approved under this plan. The fifth harvest occurred in 2001 and was the first entry in that region since the clear-cut in the 1880s (California Department of Forestry and Fire Protection 2001). Other selective harvests completed include one in the Southern Unit, which was entered in 1996, as well as one harvest in the Late Successional Unit, entered in 2007 for the first time since the clear-cut. Unlike the other management units at Byrne-Milliron Forest, this harvest area has different management goals, which include minimal logging and an overall reduction in disturbance in an effort to promote late successional features and maintain habitat features such as snags and LWD. All harvests completed under THPs had subsequent plantings following the logging operations. Primary goals for the forest initially set forth
1 CU = Central Unit; SU = Southern Unit; LSU = Late Successional Unit
2 S = skyline cable; T = tractor
3 Most recent harvest
22
by Carlton Byrne and since upheld by the Land Trust include sustainable forestry, recreation and natural resource protection.
Study Design
Data collection for the study began in May 2012 and ended in July 2013. Site selection was determined by analyzing ground accessibility, timber harvest maps and applicable THPs and NTMPs. The replicated sample design consisted of five second-growth sites that had undergone selective-harvest since the property was acquired by the Land Trust of Santa Cruz County in 1984. Twenty 0.032 hectare (20 m diameter) sample plots were randomly selected within each of the five sites (Figure 4). All plot locations were situated at least 10 m from sensitive habitats and 200 m from main access roads (Russell and Michels 2010). Plot locations were selected at random with the use of a random number generator and coordinates were recorded with a Garmin GPS device.
Each plot was further divided into sample quadrants to determine relative herbaceous cover.
23
Figure 4. Site Schematic and 20-Meter Diameter Sample Plot. Image by Derrick Wynes.
Data Collection
Physiographic variables recorded at each plot included slope, aspect and
elevation. All tree species >10 cm were counted and measured using a DBH (diameter at breast height) tape. Seedlings and sprouts were also tallied, identified and categorized by whether or not they exceeded one meter in height. Canopy cover was determined using a convex spherical densiometer, with readings taken in each of the four cardinal directions.
LWD was also counted and circumference and length were recorded for each occurrence.
Herbaceous species within each plot were identified and visual estimates were made to determine relative species composition. In the event of an unidentifiable plant, a sample of the specimen was collected for later identification using the Jepson Manual (Hickman 1993).
24 Analytical Methods
A Pearson’s product-moment correlation matrix was used to analyze relationships between the independent variables (years since last harvest, number of harvest re-entries and the percentage cut per hectare) and dependent variables (aspect, elevation, slope, canopy cover, stand density, size classes of coast redwood and tanoak, size and number of LWD, native and exotic species richness, coast redwood associated species, percentage of broad leaved helleborine, forget-me-not and English holly, as well as basal area and dominance of coast redwood). To look at stem size variation in further depth, coast redwood and tanoak were divided into seven different size classes: <10 cm, 10-24 cm, 25-49 cm, 50-99 cm, 100-149 cm, 150-199 cm and >200 cm DBH (Giusti 2007). LWD was also broken up into size classes including short (2-8 m), medium (9-15 m) and long (>15 m) (Sinclair 2013). In addition, native, exotic and coast redwood associated herbaceous species were examined to determine species prevalence and richness at each site. One-way analysis of variance (ANOVA) were conducted to determine whether there were significant differences between each of the earlier mentioned dependent variables and the number of harvests, percentage harvested and date of last harvest re-entry. Post hoc analyses used the Bonferroni correction. All statistical analyses were conducted using IBM SPSS Statistics 21 and Microsoft Office Excel 2013.
RESULTS
There were a number of variations in forest structure and composition among study sites, with significant differences determined for canopy cover, stand density, density of LWD, coast redwood sprouts, coast redwood associated herbaceous species
25
and exotic species richness. Physiographic variables including aspect, canopy cover, elevation and slope were relatively similar among sites (Table 2). The median aspect was 42° in the 1987/2004 site, 71° in the 1990/2007 site, 90° in the 1996 site, 68° in the 2001 site and 84° in the 2007 site, respectively.
Table 2. Physiographic characteristics among study sites.
Physiographic 1987/2004 1990/2007 1996 2001 2007 Variables Mean S.E. Mean S.E. Mean S.E. Mean S.E. Mean S.E.
Canopy cover 0.97 0.004 0.98 0.002 0.93 0.02 0.98 0.004 0.98 0.002 Elevation 1213 25.42 928 19.68 1071 20.68 1475 21.46 1094 30.24 Slope (%) 41.9 2.97 36.05 1.72 39 2.84 41.6 2.62 39.8 3.18
Canopy Cover
The 2007 site had the highest canopy cover, followed by the 1990/2007 site, the 2001 site, the 1987/2004 site and the 1996 site (Figure 5). When years since harvest was analyzed, ANOVA indicated that the 1996 site was significantly different from all other harvest sites including the 2007 site (p = <0.001), the 1987/2004 site (p = 0.006) and the 2001 site (p = 0.002); a Pearson’s product-moment correlation coefficient indicated this was a weak negative relationship (r = -0.04). Analysis of canopy cover and the
percentage cut per hectare also indicated that the 1996 harvest site was significantly different from all other sites including the 2007 site (p = 0.001), the 1987/2004 site (p = 0.01), the 2001 site (p = 0.004) and the 1990/2007 site (p = 0.004). Again, this was a weak negative relationship (r = -0.04). Analysis of canopy cover and number of
harvest re-entries indicated there was not a significant difference among sites (p = 0.2).
26
Figure 5. Mean canopy cover among study sites with 95% confidence intervals (sites with the same letter were not significantly different from one another).
Stand Density
Mean species density varied among study sites (Table 3). Species identified at the Byrne-Milliron Forest included coast redwood, tanoak (Lithocarpus densiflorus), Pacific madrone, Douglas-fir and Quercus ssp., which were determined to either be coast live oak (Quercus agrifolia) or interior live oak (Quercus wislizenii) (Figure 6). The 1990/2007 harvest site had the highest stand density of all sites, followed by the 1987/2004 site, the 1996 site, the 2007 site and the 2001 site (Figure 7).
0.93
27 Table 3. Mean stand density among study sites.
1987/2004 1990/2007 1996 2001 2007
Stand Density Mean S.E. Mean S.E. Mean S.E. Mean S.E. Mean S.E.
All species 11.9 0.96 12.85 1.41 11.15 0.98 5.95 1.06 7.05 0.68 Coast redwood 10.1 0.81 11.25 1.24 9.1 1.09 8.85 1.04 6.55 0.69 Tanoak 49.15 11.49 69.15 14.67 49.65 8.61 25.95 5.78 16.85 2.69 Pacific madrone 0.6 0.5 0.1 0.07 0.3 0.15 3.1 1.44 1.25 0.95 Oak 2.35 1.25 4.6 1.22 5.4 3.12 0.55 0.26 0.65 0.25 Douglas-fir 0.4 0.28 0.4 0.28 0 0 0.6 0.29 0 0 Big-leaf maple 0.05 0.05 0.05 0.05 0.4 0.35 0 0 0.35 0.3
Figure 6. Frequency of tree species among study sites.
0 50 100 150 200 250
2007 2001 1996 1987/2004 1990/2007
Stand Density
Sites
Coast Redwood Tanoak Madrone Oak Douglas-Fir
28
Figure 7. Mean stand density among study sites with 95% confidence intervals (sites with the same letter were not significantly different from one another).
When ANOVA was used to analyze stand density and years since last harvest, results showed there was not a significant difference among study sites (p = 0.052).
However, when analyzed with number of harvest re-entries, there was a significant difference between one and two re-entries (p = 0.001); this was a positive relationship according to Pearson’s product-moment correlation coefficient (r = 0.15). When stand density was analyzed with the percentage cut per hectare, the 2007 site had a significantly lower density than the 1990/2007 site (p = 0.001). Stand density had a positive
relationship with this independent variable (r = 0.27).
In addition to having the highest mean stand density, the 1990/2007 site also had the highest mean density of coast redwood, followed by the 1987/2004 site, the 1996 site, the 2001 site and the 2007 site. When coast redwood density was analyzed with years since last harvest, a one-way ANOVA indicated that the difference among sites was not
5.95 7.05
29
significant (p = 0.8). When analyzed with number of harvest re-entries, there was a significant difference between sites re-entered once versus sites re-entered twice (p = 0.007). A Pearson’s product-moment correlation coefficient indicated this
relationship was positive (r = 0.1). There were also significant findings for coast
redwood density and the percentage cut per hectare, with results indicating that the 2007 site had a significantly lower density of coast redwood in comparison with the 1990/2007 site (p = 0.012); this relationship was positive (r = 0.26).
Density of tanoak varied across sites. The 1990/2007 site had the highest mean, followed by the 1996 site, the 1987/2004 site, the 2001 site and the 2007 site. Mean tanoak density was determined to have a significant positive relationship with years since harvest (p = 0.049; r = 0.15), but was not significantly different among sites when analyzed with number of harvest re-entries (p = 0.32) or the percentage cut per hectare (p = 0.076).
Other observations included Pacific madrone, oak, Douglas-fir and big leaf maple. Density of Pacific madrone was highest in the 2001 site, followed by the 2007 site, the 1987/2004 site, the 1996 site and the 1990/2007 site. Mean density of oak was highest in the 1996 site, followed by the 1990/2007 site, the 1987/2004 site, the 2007 site and the 2001 site. Douglas-fir was scarce among sites, with the 2001 site having the highest mean density, followed by the 1990/2007 site and the 1987/2004 site. There were no observations of this species in either the 1996 or 2007 sites. Big leaf maple was also not very prevalent, with a mean <0.50 for all sites. The 1996 site had the highest density
30
of big-leaf maple, followed by the 2007 site, the 1987/2004 site and the 1990/2007 site;
there were no observations in the 2001 site.
Coast Redwood Size Classes
Coast redwood size class distribution varied among sites (Figure 8; Table 4). The 1987/2004 site had the highest number of stems <10 cm, followed by the 1996 site, the 2001 site, the 1990/2007 site and the 2007 site. When analyzed with years since harvest, a one-way ANOVA indicated a significant difference between the 1987/2004 site in comparison with the 2007 site (p = 0.002) and the 2001 site (p = 0.046). A Pearson’s product-moment correlation coefficient indicated a positive relationship between years since last harvest and coast redwoods <10 cm (r = 0.18).
Although there was not a significant difference among sites for number of harvest re-entries and stems <10 cm (p = 0.19), there were significant findings for the percentage cut per hectare and coast redwood in this size class. The 1987/2004 site had a
significantly higher number of stems in comparison with the 2007 site (p = 0.008) and the 1990/2007 site (p = 0.03); this relationship was negative (r = -0.05).
The 1990/2007 site had the highest number of coast redwood 10-24 cm, followed by the 1987/2004 site, the 2001 site, the 1996 site and the 2007 site. A one-way ANOVA indicated there was not a significant difference among sites for stems in this size class and years since last harvest (p = 0.15). However, when analyzed with number of harvest re-entries, stems 10-24 cm were significantly lower in sites with one re-entry compared with two re-entries (p = <0.001); this relationship was positive (r = 0.22). The 1996 site,
31
the 2001 site and the 2007 site, which were only re-entered once, had a lower number of stems 10-24 cm in comparison with the sites re-entered twice.
When ANOVA was used to analyze the percentage cut per hectare, the mean number of stems 10-24 cm in the 2007 site was significantly lower than the 1987/2004 site (p = <0.001), the 1990/2007 site (p = <0.001), the 1996 site (p = 0.046) and the 2001 site (p = 0.038). A Pearson’s product-moment correlation coefficient indicated this relationship was positive (r = 0.36).
The 1990/2007 site had the highest mean number of coast redwood 25-49 cm, followed by the 1987/2004 site, the 1996 site, the 2001 site and the 2007 site. When ANOVA was used to analyze this size class with years since last harvest, there was not a significant difference among sites (p = 0.069). There was also not a significant
difference among sites for number of harvest re-entries (p = 0.35) or the percentage cut per hectare (p = 0.74).
The 1990/2007 site had the highest mean number of coast redwood 50-99 cm, followed by the 1996 site, the 2001 site, the 2007 site and the 1987/2004 site. There was not a significant difference for coast redwood stems in this size class when analyzed with years since last harvest (p =0.71), number of harvest re-entries (p = 0.42) or the
percentage cut per hectare (p = 0.37).
The 1987/2004 site had the highest mean number of stems 100-149 cm, followed by the 1996 site, the 2007 site, the 2001 site and the 1990/2007 site. There were not significant findings for coast redwood stems in this size class when analyzed with years
32
since last harvest (p = 0.76), number of harvest re-entries (p = 0.55) or the percentage cut per hectare (p = 0.71).
The 2007 site had the highest mean number of coast redwood 150-199 cm,
followed by the 1987/2004 site and the 2001 site. There were no observations of this size class found in the 1996 or 1990/2007 sites. When analyzed with years since last harvest, results indicated there was not a significant difference among sites (p = 0.16). There was also not a significant difference among sites for number of harvest re-entries (p = 0.09).
However, there were significant findings for this size class when analyzed with the percentage cut per hectare. The 2007 site had a significantly larger number of stems in comparison with the 1990/2007 site (p = 0.002), the 1996 site (p = 0.002), the 1987/2004 site (p = 0.018) and the 2001 site (p = 0.018); this was a negative relationship (r = -0.35).
There were only a small number of occurrences where stems exceeded 200 cm.
This included one occurrence in the 2001 site and one occurrence in the 1990/2007 site.
When ANOVA was used to analyze stems in this size class with years since last harvest, these occurrences were not found to be significant (p = 0.63). Results for stems >200 cm and number of harvest re-entries were also not significant (p = 0.78). In addition, there was not a significant difference found when analyzed with the percentage cut per hectare (p = 0.56).
33
Figure 8. Coast redwood size class distribution among study sites.
Table 4. Means and standard errors for coast redwood size classes.
Coast Redwood
Tanoak size classes varied among sites, with stems <10 cm occurring with the highest prevalence (Table 5). The 1990/2007 site had the highest number of stems in this size class, followed by the 1996 site, the 1987/2004 site, the 2001 site and the 2007 site.
When ANOVA was used to analyze years since last harvest, there was not a significant 0
<10cm 10-24cm 25-49cm 50-99cm 100-149cm 150-199cm >200cm
34
difference among sites (p = 0.21). However, there were significant findings for tanoak stems <10 cm and number of harvest re-entries (p = 0.004). A Pearson’s
product-moment correlation coefficient indicated this was a weak positive relationship (r = 0.05).
When tanoak <10 cm was analyzed with the percentage cut, there was a significant difference between the 1990/2007 site in comparison with the 2007 site (p = 0.004) and the 2001 site (p = 0.009); this relationship was positive (r = 0.24).
Tanoak 10-24 cm had the highest density in the 1996 harvest site, followed by the 1990/2007 site, the 1987/2004 site and the 2007 site. There were no observations of this size class present in the 2001 site. When analyzed with years since last harvest, a one-way ANOVA indicated there was a significant difference between the 2001 site and the 1996 site (p = 0.006); this relationship was positive (r =0.21). There was not a
significant difference for tanoak 10-24 cm and number of harvest re-entries (p= 0.56).
However, there was a significant difference among sites when this size class was
analyzed with the percentage cut per hectare. This was a positive relationship (r = 0.12).
analyzed with the percentage cut per hectare. This was a positive relationship (r = 0.12).